Kube-green Auto-Shutdown K8s pods when not in use

                                          

Today, we will explore an interesting K8s plugin called 'kube-green' that can help scale down or up the pods as needed during working hours/weekends. Once the initial configuration is complete, this plugin will automatically manage it.

Kube-green: This Kubernetes(k8s) operator enables the shutdown of environments or specific resources, allowing for optimal resource utilisation and minimising energy waste. It helps to bring up/down deployments and cronjobs.

Requirements:

K8s cluster (min. version 1.19) (I Am using version 1.25.4)

Step 1: Install kube-green,

Clone the below repo which has all the configuration details,

git clone https://github.com/DevOpsArts/devopsart-kubegreen.git

cd devopsart-kubegreen

Install cert-manager,

kubectl apply -f cert-manager.yaml

Install kube-green,

kubectl apply -f kube-green.yaml

Get all resources related to kube green by,

kubectl get all -n kube-green

Step 2: Deploy Nginx web for validation using helm,

helm repo add bitnami https://charts.bitnami.com/bitnami

helm install nginx-web bitnami/nginx

kubectl get pods

Now Nginx pod is up and running in the k8s cluster

Step 3: Configure kube-green 

We will scale down and up the nginx web using kube-green,

Go to the same git cloned folder and update the timing and deployment namespace details,

https://github.com/DevOpsArts/devopsart-kubegreen/blob/main/working-hours.yml

cd devopsart-kubegreen

cat working-hours.yml

 apiVersion: kube-green.com/v1alpha1

kind: SleepInfo

metadata:

  name: working-hours

  namespace: default

spec:

  weekdays: "1-5"

  sleepAt: "08:40"

  wakeUpAt: "08:42"

  timeZone: "Etc/UTC"

Update the bold letters according to your requirements, and it will scale down and up all the deployments in the cluster.

The screenshot below shows how it worked. The pod was scheduled to scale down at 8:40 AM UTC and scale up at 8:42 AM UTC, and it scaled down at 8:40 AM UTC according to the configuration.

And at 8.42AM UTC, the pod came up as per the configuration,

Below configuration will help to scale down the pods with exceptions.

https://github.com/DevOpsArts/devopsart-kubegreen/blob/main/working-hours-expection.yml

We can check the kube-green logs from kube-green pods for scale down/up status,

kubectl get pods -n kube-green

kubectl logs -f kube-green-controller-manager-6446b47f7c-hbmtx -n kube-green

Step 4: Kube-green monitoring,

We can monitor the resource utilization of the Kube Green resources and check the status of pod scale-down and scale-up. Kube Green exposes Prometheus metrics on port number 8080 with the /metrics path. We can configure it in Grafana to monitor the status.

That's all. We have deployed Kube-Green in the K8s cluster and validated it by scaling down and up.

Reference,

https://kube-green.dev/docs/